Neural networks (NN) and support vector machines (SVM) play key roles in machine learning and data analysis. However, it is known that these popular learning techniques face some challenging issues such as: intensive human intervene, slow learning speed, poor learning scalability. The newly proposed Extreme Learning Machine (ELM) can resolve those challenging issues. This talk will give the answers on the reasons: 1) why ELM can work as universal approximator and why tuning is not required; 2) why ELM can differentiate any disjoint classification regions; and 3) why ELM outperforms LS-SVM. This talk will also introduce an efficient incremental implementation of ELM and discuss some open problems.
Uploaded by chris85 on